An adaptive deployment method and system for edge intelligent networks in discrete manufacturing

Abstract

This invention relates to an adaptive deployment method and system for edge intelligent networks in discrete manufacturing. The method aims to maximize network throughput, minimize transmission latency, and maximize network reliability. It establishes a QoS-driven edge intelligent network deployment optimization model to ensure that the network transmission requirements of all devices in the factory, especially critical equipment, are met. It employs improved K-means and particle swarm optimization algorithms, combined with dynamic clustering algorithms and adaptive weight adjustment mechanisms, to effectively improve the data rate of intelligent devices and network reliability, while reducing transmission latency and system interference. This invention can adapt to factories of different sizes and complex environments, optimize network performance, significantly improve the efficiency of edge intelligent deployment, and provide a stable and reliable communication solution for discrete manufacturing smart factories and intelligent manufacturing in the Industry 5.0 era.

Description

Technical Field

[0001] This invention belongs to the field of industrial communication network technology, and specifically relates to an adaptive deployment method and system for edge intelligent networks oriented towards discrete manufacturing. Background Technology

[0002] With the development of industrial automation and intelligent manufacturing, factory environments are becoming increasingly complex, leading to a growing demand for wireless connectivity between equipment and systems. Discrete manufacturing, in particular, requires a network system that can adapt to frequent changes in layout and equipment configuration, while also meeting increasingly stringent quality of service requirements, including low latency, high throughput, and high reliability. However, traditional network deployment strategies often struggle to meet these demands, especially in constantly evolving manufacturing environments.

[0003] In existing technologies, network deployment typically employs static design schemes. These schemes often overlook dynamic changes within the factory, such as frequent adjustments to factory layouts and rapid equipment movement. This leads to problems in actual operation, including unstable connections, increased data transmission latency, and the inability to update configurations in a timely manner. Furthermore, traditional optimization algorithms often fail to adequately consider the dynamic changes in real-time data, resulting in inflexible network adjustments that cannot effectively address the impact of equipment movement and task changes within the factory. This can potentially lead to decreased production efficiency and economic losses. Summary of the Invention

[0004] To address the aforementioned challenges, this invention develops an adaptive deployment method and system for edge intelligent networks in discrete manufacturing. This system not only adapts to dynamic changes and uncertainties within the factory but also ensures that the network performance requirements of all equipment, especially critical equipment, are met. Furthermore, by combining modern communication technologies and intelligent optimization algorithms, this invention not only enhances the data processing capabilities and response speed of factory equipment but also significantly reduces production interruptions or quality issues caused by network problems, thereby ensuring efficient communication in the production process and the stability of factory operations.

[0005] The purpose of this invention is to provide an adaptive deployment method and system for edge intelligent networks in discrete manufacturing, aiming to address the complex dynamic changes and uncertainties within discrete manufacturing plants and ensure that network reliability and quality of service meet requirements under any circumstances. By constructing an edge intelligent network deployment optimization model driven by Quality of Service (QoS) under limited spectrum resources, the edge intelligent network deployment possesses greater flexibility and adaptability. By jointly using an improved K-means clustering algorithm and a Particle Swarm Optimization (PSO) algorithm, the edge intelligent network deployment strategy can quickly converge to a high-quality network configuration solution in a shorter time, improving network deployment efficiency and network stability.

[0006] The objective of this invention is achieved through the following technical solution:

[0007] An adaptive edge intelligent network system for discrete manufacturing is disclosed. The system includes multiple factory devices, multiple heterogeneous cellular network base stations, multiple mobile edge computing (MEC) servers, and a cloud server. The heterogeneous cellular network base stations include macro base stations and micro base stations. Micro base stations are deployed in areas where macro base station signal coverage is weak or difficult to achieve, and are connected to macro base stations via wireless fronthaul technology to enhance network signal quality and coverage. Each factory device connects to an edge intelligent network node that provides the maximum signal-to-interference-plus-noise ratio (SINR) based on its actual communication environment, ensuring optimal network connection quality at all times. Each macro base station connects to a mobile edge server, which is responsible for local data processing and computation tasks, reducing data transmission latency and improving edge computing efficiency. Ultimately, the cloud server enables coordinated and optimized management of global data.

[0008] An adaptive deployment method for edge intelligent networks in discrete manufacturing, comprising the following steps:

[0009] (1) For the above system, with the goal of reducing system interference and maximizing network service quality, an adaptive deployment optimization problem of edge intelligent network nodes is constructed based on the current network conditions and the network requirements of factory equipment.

[0010] (2) The improved K-means clustering algorithm and PSO algorithm are used to solve the adaptive deployment optimization problem, and the intelligent deployment and dynamic adjustment of all heterogeneous cellular network base stations and all mobile edge servers are completed, thereby improving the quality of network services and the utilization rate of computing resources.

[0011] Preferably, the specific steps for constructing the adaptive deployment optimization problem of edge intelligent network nodes in step (1) above are as follows:

[0012] (1-1) Assume the system uses frequency division multiple access (FDMA) technology, macro base stations and micro base stations operate on different frequency bands, and allocate the required network resources to factory equipment through a collaborative approach. a ij =1 indicates that the i-th factory device is associated with the j-th edge intelligent network node, and vice versa, i∈[1,D]; x ik =1 indicates that the k-th subcarrier is assigned to the i-th factory device, otherwise it is not assigned, k∈[1,K];

[0013] (1-2) Construct a QoS model for a discrete manufacturing edge intelligent network system; considering the presence of numerous obstacles such as metal surfaces, large equipment, and buildings in the factory, and the impact of machine operation and production activities on the transmission quality of wireless signals, the wireless channel between factory device i and edge intelligent network node j is described using the WINNER II B3 channel model; assuming the transmit power of factory device i is... The carrier frequency is f ik The distance between device i and its associated edge intelligent network node j is d. ij The path loss model is then expressed as:

[0014]

[0015] Where LOS and NLOS represent line-of-sight propagation and non-line-of-sight propagation, respectively, X σ Let be the loss correction factor caused by obstacles; therefore, the signal power received by edge intelligent network node j is expressed as: Affected by shadow decay, For a random variable that follows a log-normal distribution, the signal reception failure rate is defined as the received power being less than the minimum received power. The probability of that, that is:

[0016]

[0017] in, Let be the average signal power received by base station j, σ be the standard deviation of shadow fading, and Q(z) be the probability that the random variable x, which follows a standard normal distribution, is greater than z.

[0018]

[0019] Considering the co-channel interference from other devices using the same carrier frequency as device i within the factory, as well as the impact of mechanical vibration, electromagnetic interference, and random noise generated by equipment operation on the data transmission of device i, the uplink SINR of the i-th factory device is expressed as:

[0020]

[0021] in, This indicates the co-channel interference experienced by device i. P represents the thermal noise of factory equipment. ijk (t)=A∑ l δ(t-lT) represents the noise generated by equipment operation, switching action or other periodic activities, where A is the pulse amplitude, δ(t) is the Dirac delta function, T is the pulse period, and lT is the specific time point when the l-th pulse occurs;

[0022] Assume m ij b represents the data size transmitted from the i-th factory device to the j-th edge intelligent network node. ij Given the network bandwidth allocated to the i-th factory device, the sum of transmission delays for all factory devices within the coverage area of ​​the j-th edge intelligent network node is expressed as:

[0023]

[0024] Assuming the j-th edge intelligent network node has sufficient network capacity, the sum of the data size of all factory devices within its coverage area is expressed as:

[0025]

[0026] The system employs quadrature phase shift keying (QPS) modulation for signal transmission to effectively utilize limited communication bandwidth. If the m-th factory device... ij The probability of successful data transmission for each bit or data unit is the product of its success rate. Therefore, the probability that all factory equipment within the coverage area of ​​the j-th edge intelligent network node successfully transmits data is:

[0027]

[0028] (1-3) Constructing an adaptive deployment optimization problem for a discrete manufacturing edge intelligent network: If the key indicators of system network service quality include communication reachability, system latency, and transmission reliability, then the service quality of edge intelligent network nodes is defined as follows:

[0029]

[0030] Where z represents the deployment location of the edge intelligent network node, ω1,ω2,ω3∈[0,1] are the weighting coefficients of each key indicator in the system, and ω1+ω2+ω3=1, T j T is the sum of the actual data upload latency of all factory equipment within the coverage area of ​​the j-th edge intelligent network node. req The minimum latency that should be met for data transmission to factory equipment within its coverage area; R j R represents the transmission capacity that the j-th edge intelligent network node can provide. req Average network capacity required for data upload from factory equipment within the coverage area; PSR j To ensure the reliability of data transmission between factory equipment within the coverage area of ​​the j-th edge intelligent network node, PSR req The minimum reliability that must be met for successful data transmission from factory equipment within the coverage area;

[0031] Considering the uneven distribution of factory equipment and communication needs, the rational and efficient deployment of edge intelligent network nodes can provide factory equipment with low latency, high reliability, and high data rate quality of service, while meeting the constraints of maximum system capacity and minimum transmission latency. Therefore, the adaptive deployment optimization problem of edge intelligent network nodes can be expressed as:

[0032]

[0033] C7:ω1+ω2+ω3=1

[0034] in, This is a service quality redundancy factor designed to cope with service fluctuations such as increased network load. Constraint C1 indicates that a device will only connect to one base station at a time. Constraint C2 indicates that each device transmits data only on a specific sub-channel, and data transmission through multiple sub-channels is not allowed at the same time, thus avoiding inter-channel interference and packet collisions. Constraint C3 indicates that the total bandwidth allocated to all devices cannot exceed the total bandwidth of the base station, and the bandwidth allocated to each device must be non-negative. Constraints C4 to C6 indicate that the uplink transmission rate, latency, and reliability of each factory device i connected to the edge intelligent network node j can be basically guaranteed, where R min T max and PSR min These represent the minimum transmission rate, maximum allowable transmission delay, and minimum reliability of factory equipment i, respectively.

[0035] Preferably, the specific steps of step (2) above are as follows:

[0036] (2-1) Identify potential QoS-deficient areas

[0037] By monitoring the real-time data transmission of various devices in a discrete manufacturing smart factory, and utilizing a preset QoS threshold standard R... min T max and PSR min Devices with unmet QoS requirements are detected, and then the number and spatial distribution of these devices are analyzed to quickly identify potential underserved areas. The set of devices with unmet QoS requirements is...

[0038] (2-2) The improved K-means algorithm is used to perform dynamic clustering initialization on factory equipment that does not meet QoS, and the cluster center is used as the candidate deployment location of edge intelligent network nodes.

[0039] Assume x j For decision variables, j indexes all possible micro base station locations, and x... j=1 indicates that a base station is deployed at location j, otherwise it is not deployed. For each factory device i∈U, a ij =1 indicates that device i, whose QoS is not satisfied, is covered by the micro base station at location j; conversely, 0 indicates that the QoS of device i is still not satisfied. If the deployment cost of each micro base station is C... j The objective function of dynamic clustering is then expressed as:

[0040]

[0041] For the optimization problem P2, the improved K-means algorithm is used to solve it. The number of cluster centers is dynamically adjusted in real time according to the number of devices whose QoS is not met and their spatial distribution, until an edge intelligent network node deployment scheme that can both ensure the service quality requirements of all devices in set U and maximize cost-effectiveness is found.

[0042] (2-3) Use the adaptive weighted PSO algorithm to accurately adjust the location and configuration of heterogeneous cellular network base stations and MEC servers;

[0043] In the PSO algorithm, each particle has a position vector X and a velocity vector V. X represents the particle's position in the solution space, and V represents the direction and velocity that determine its flight. Assume V... i t This represents the velocity of the i-th particle at time t. This represents the position of the i-th particle at time t. G represents the best position found in the history of the i-th particle. best This represents the best position found throughout the entire population's history. In each iteration, each particle is compared to its own historical best position. and the global best position g in the group best To update its speed and position, that is:

[0044]

[0045]

[0046] Where r1 and r2 are random numbers, and c1 and c2 represent the acceleration weights that push the particle to its individual optimal position and the group optimal position, respectively. To maintain a balance between convergence speed and search effect, neither c1 nor c2 is 0; ω is the inertia weight coefficient at time t, which controls the continuity of particle velocity updates. A higher weight coefficient is helpful for global search, while a lower weight coefficient is helpful for local search. Fixed inertia weights cannot effectively balance exploration and utilization in complex multi-objective optimization problems, causing particles to converge to local optima too early and making it difficult to adapt to the dynamically changing factory environment. To effectively balance global and local search, the inertia weight is defined as:

[0047] ω=ω max -(ω max -ω min (t / T) p

[0048] Where, ω max and ω min represents the maximum and minimum values ​​of the inertia weight coefficient, respectively. p is a constant, and T is the total number of iterations of the algorithm. In the early stage of the search, a larger inertia weight is used to help the algorithm explore a wider solution space. In the later stage of the iteration, the inertia weight is gradually reduced, which enhances the local search ability of the particles, improves the accuracy of the optimal solution and the convergence of the algorithm.

[0049] The beneficial effects of this invention include:

[0050] 1. Improve network coverage quality and stability: By adopting a combination of macro base stations and micro base stations, and using wireless fronthaul technology to connect micro base stations and macro base stations, the network service quality in areas with weak or difficult signal coverage in discrete manufacturing smart factories can be effectively enhanced, ensuring that factory equipment always maintains a stable network connection.

[0051] 2. Optimize network resource utilization efficiency: By using improved K-means clustering and PSO algorithms, the deployment of heterogeneous cellular network base stations and MEC servers can be dynamically adjusted. This can optimize resource allocation according to the actual communication environment and equipment requirements, thereby improving computing resource utilization efficiency and network service quality.

[0052] 3. Reduce data transmission latency and improve edge computing efficiency: Through the collaboration between each macro base station and the local MEC server, this invention can distribute data processing tasks to edge computing nodes, reduce data transmission distance and latency, improve real-time processing capabilities, and finally perform global optimization by the cloud server to ensure the efficient operation of the entire system.

[0053] 4. Enhanced Adaptive Optimization Capability: This invention uses dynamic clustering initialization and adaptive iterative optimization mechanisms to dynamically adjust the deployment location and configuration of heterogeneous cellular network base stations and MEC servers based on devices with unmet QoS requirements. This ensures that the system can quickly respond to changes in the factory environment, achieve a more efficient and low-interference operating state, and meet the dynamic needs of the factory. Attached Figure Description

[0054] The principles of the present invention can be more fully understood by referring to the following accompanying drawings. Obviously, the following drawings are merely some embodiments of the present invention, and those skilled in the art can create other drawings based on these drawings without any inventive effort.

[0055] Figure 1This is a schematic diagram of a discrete manufacturing edge intelligent network architecture in an exemplary embodiment of the present invention;

[0056] Figure 2 This is a flowchart illustrating the solution of the edge intelligent network deployment optimization problem using the improved K-means algorithm and the PSO algorithm in an exemplary embodiment of the present invention.

[0057] In the diagram: 1. Factory equipment, 2. Micro base station, 3. Wireless fronthaul connection, 4. Macro base station, 5. MEC server, 6. Cloud server. Detailed Implementation

[0058] Exemplary embodiments of the present invention will now be described in detail with reference to the accompanying drawings. However, the embodiments described with reference to the drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention. Furthermore, the drawings are merely illustrative diagrams of the present invention and are not drawn to scale; some block diagrams in the drawings represent functional entities, but do not necessarily correspond to physically or logically independent entities. Functional entities may be implemented in software, or in one or more hardware modules or integrated circuits, or in different networks and / or processors and / or microcontrollers.

[0059] Example 1:

[0060] An adaptive edge intelligent network system for discrete manufacturing is provided, such as... Figure 1 As shown, the system includes multiple factory devices 1, multiple micro base stations 2, multiple macro base stations 4, multiple MEC servers 5, and a cloud server 6. Micro base stations 2 are primarily deployed in areas where macro base station 4 has weak or difficult-to-cover signal coverage, and are connected to macro base stations 4 via wireless fronthaul technology 3, thereby enhancing the overall network signal quality and coverage. Factory devices 1 are selected to provide the highest SINR edge intelligent network node access, ensuring that the devices always maintain optimal network connection quality. Each macro base station 4 is connected to an MEC server 5, responsible for distributing local data processing and computing tasks, reducing data transmission latency and improving edge computing efficiency. Finally, the cloud server 6 performs global data coordination and optimization management.

[0061] Example 2:

[0062] For the system architecture of Example 1, an adaptive deployment method for edge intelligent networks for discrete manufacturing is provided, the steps of which are as follows:

[0063] S1: With the goal of reducing system interference and maximizing network service quality, based on the current network conditions and the network requirements of factory equipment, construct an adaptive deployment optimization problem for edge intelligent network nodes (including heterogeneous cellular network base stations and MEC servers), including the following sub-steps:

[0064] Sub-step S11: Assume the system uses frequency division multiple access (FDMA) technology, macro base station 4 and micro base station 2 operate on different frequency bands, and allocate the required network resources to the factory equipment through a cooperative approach. ij =1 indicates that the i-th factory device is associated with the j-th edge intelligent network node, and vice versa, i∈[1,D]; x ik =1 indicates that the k-th subcarrier is assigned to the i-th factory device, otherwise it is not assigned, k∈[1,K].

[0065] Sub-step S12: Construct the QoS model for the discrete manufacturing edge intelligent network system. Considering the presence of numerous obstacles in the factory, such as large metal surfaces, large equipment, and buildings, and the impact of machine operation and production activities on the transmission quality of wireless signals, the wireless channel between factory device i and edge intelligent network node j is described using the WINNER II B3 channel model. Assume the transmit power of factory device i is... The carrier frequency is f ik The distance between device i and its associated edge intelligent network node j is d. ij The path loss model is then expressed as:

[0066]

[0067] Where LOS and NLOS represent line-of-sight propagation and non-line-of-sight propagation, respectively, X σ Let $\frac{ ... Affected by shadow decay, For a random variable that follows a log-normal distribution, the signal reception failure rate is defined as the received power being less than the minimum received power. The probability of that, that is:

[0068]

[0069] in, Let be the average signal power received by base station j, σ be the standard deviation of shadow fading, and Q(z) be the probability that the random variable x, which follows a standard normal distribution, is greater than z.

[0070]

[0071] Considering the co-channel interference from other devices using the same carrier frequency as device i within the factory, as well as the impact of mechanical vibration, electromagnetic interference, and random noise generated by equipment operation on the data transmission of device i, the uplink SINR of the i-th factory device can be expressed as:

[0072]

[0073] in, This indicates the co-channel interference experienced by device i. P represents the thermal noise of factory equipment. ijk (t)=A∑ l δ(t-lT) represents the noise generated by equipment operation, switching action or other periodic activities, where A is the pulse amplitude, δ(t) is the Dirac delta function, T is the pulse period, and lT is the specific time point when the l-th pulse occurs.

[0074] Assume m ij b represents the data size transmitted from the i-th factory device to the j-th edge intelligent network node. ij Let the network bandwidth allocated to the i-th factory device be denoted as . Then, the sum of the transmission delays of all factory devices within the coverage area of ​​the j-th edge intelligent network node is expressed as:

[0075]

[0076] Assuming the j-th edge intelligent network node has sufficient network capacity, the sum of the data size of all factory devices within its coverage area is expressed as:

[0077]

[0078] The system employs quadrature phase-shift keying (QPS) modulation for signal transmission to effectively utilize limited communication bandwidth. If the m-th factory device... ij The probability of successful data transmission for each bit or data unit is the product of its success rate. Therefore, the probability that all factory equipment within the coverage area of ​​the j-th edge intelligent network node successfully transmits data is:

[0079]

[0080] Sub-step S13: Constructing an adaptive deployment optimization problem for a discrete manufacturing edge intelligent network. If the key indicators of system network service quality include communication reachability, system latency, and transmission reliability, then the service quality of edge intelligent network nodes is defined as follows:

[0081]

[0082] Where z represents the deployment location of the edge intelligent network node, ω1,ω2,ω3∈[0,1] are the weighting coefficients of each key indicator in the system, and ω1+ω2+ω3=1. T j T is the sum of the actual data upload latency of all factory equipment within the coverage area of ​​the j-th edge intelligent network node. req The minimum latency that should be met for data transmission to factory equipment within its coverage area; R jR represents the transmission capacity that the j-th edge intelligent network node can provide. req Average network capacity required for data upload from factory equipment within the coverage area; PSR j To ensure the reliability of data transmission between factory equipment within the coverage area of ​​the j-th edge intelligent network node, PSR req The minimum reliability that must be met for successful data transmission from factory equipment within the coverage area.

[0083] Considering the uneven distribution of factory equipment and communication needs, the rational and efficient deployment of edge intelligent network nodes can provide factory equipment with low latency, high reliability, and high data rate service quality, while meeting the constraints of maximum system capacity and minimum transmission latency. Therefore, the adaptive deployment optimization problem of edge intelligent network nodes can be expressed as:

[0084]

[0085] C7:ω1+ω2+ω3=1

[0086] in, This is a service quality redundancy factor designed to cope with service fluctuations such as increased network load. Constraint C1 states that a device will only connect to one base station at a time. Constraint C2 states that each device transmits data only on a specific sub-channel, and data transmission through multiple sub-channels is not allowed simultaneously, thus avoiding inter-channel interference and packet collisions. Constraint C3 states that the total bandwidth allocated to all devices cannot exceed the total bandwidth of the base station, and the bandwidth allocated to each device must be non-negative. Constraints C4 to C6 respectively indicate that the uplink transmission rate, latency, and reliability of each factory device i connected to the edge intelligent network node j can be basically guaranteed, where R... min T max and PSR min These represent the minimum transmission rate, maximum allowable transmission delay, and minimum reliability of factory equipment i, respectively.

[0087] Step S2: As Figure 2 As shown, an improved K-means clustering algorithm and the PSO algorithm are used to solve the adaptive deployment optimization problem, enabling intelligent deployment and dynamic adjustment of all heterogeneous cellular network base stations and all MEC servers, thereby improving network service quality and computing resource utilization. The process includes the following sub-steps:

[0088] Sub-step S21: Identify potential QoS-deficient areas. This is done by monitoring the real-time data transmission of each device in the discrete manufacturing smart factory, using a preset QoS threshold standard R. min T max and PSR minDevices with unmet QoS requirements were detected. Then, the number and spatial distribution of these devices with unmet QoS requirements were analyzed to quickly identify potential areas of insufficient service. The set of devices with unmet QoS requirements is as follows:

[0089] Sub-step S22: Utilize the improved K-means algorithm to dynamically cluster and initialize factory equipment that does not meet QoS requirements, and use the cluster centers as candidate deployment locations for edge intelligent network nodes. Assume x j For decision variables, j indexes all possible micro base station locations, and x... j =1 indicates that a base station is deployed at location j, otherwise it is not deployed. For each factory device i∈U, a ij =1 indicates that device i, whose QoS is not satisfied, is covered by the micro base station at location j; conversely, 0 indicates that the QoS of device i is still not satisfied. If the deployment cost of each micro base station is C... j The objective function of dynamic clustering is then expressed as:

[0090]

[0091]

[0092] For the optimization problem P2, an improved K-means algorithm can be used to solve it. The number of cluster centers is dynamically adjusted in real time based on the number and spatial distribution of devices with unmet QoS requirements, until a deployment scheme for edge intelligent network nodes that can both ensure the QoS requirements of all devices in set U and maximize cost-effectiveness is found. The algorithm is shown in Table 1.

[0093]

[0094]

[0095] Table 1

[0096] Sub-step S23: Utilize the adaptive weighted PSO algorithm to precisely adjust the location and configuration of heterogeneous cellular network base stations and MEC servers. In the PSO algorithm, each particle has a position vector X (the particle's position in the solution space) and a velocity vector V (determining its flight direction and velocity). Assume V... i t This represents the velocity of the i-th particle at time t. This represents the position of the i-th particle at time t. G represents the best position found in the history of the i-th particle. best This represents the best position found throughout the entire population's history. In each iteration, each particle is compared to its own historical best position. and the global best position g in the groupbest To update its speed and position, that is:

[0097]

[0098]

[0099] Where r1 and r2 are random numbers, and c1 and c2 represent the acceleration weights that push the particle to its individual optimal position and the group optimal position, respectively. To maintain a balance between convergence speed and search performance, neither c1 nor c2 is 0. ω is the inertia weight coefficient at time t, which controls the continuity of particle velocity updates. A higher weight coefficient is beneficial for global search, while a lower weight coefficient is beneficial for local search. Fixed inertia weights may not effectively balance exploration and utilization in complex multi-objective optimization problems, causing particles to converge to local optima too early and making it difficult to adapt to the dynamically changing factory environment. To effectively balance global and local search, the inertia weight is defined as:

[0100] ω=ω max -(ω max -ω min (t / T) p

[0101] Where, ω max and ω min Let represent the maximum and minimum values ​​of the inertia weight coefficient, respectively. p is a constant, and T is the total number of iterations of the algorithm. A larger inertia weight is used in the early stages of the search to help the algorithm explore a wider solution space. The inertia weight is gradually reduced in the later stages of iteration, which enhances the local search ability of the particles, improving the accuracy of the optimal solution and the convergence of the algorithm.

[0102] Table 2 shows the adaptive deployment optimization algorithm for edge intelligent networks based on the improved K-means algorithm and PSO algorithm:

[0103]

[0104]

[0105]

[0106] Table 2

[0107] It should be noted that although several units of the system for executing actions have been mentioned in the detailed description above, this division is not mandatory. In fact, according to embodiments of the present invention, the features and functions of two or more units described above can be embodied in one unit. Conversely, the features and functions of one unit described above can be further divided and embodied by multiple units. Some or all of the units can be selected to achieve the purpose of the present invention according to actual needs. Those skilled in the art can understand and implement this without any inventive effort.

[0108] Other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This invention is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein.

Claims

1. An adaptive deployment method for edge intelligent networks in discrete manufacturing, characterized in that, The steps are as follows: (1) Based on the current network conditions and the network requirements of factory equipment, construct an adaptive deployment optimization problem for edge intelligent network nodes; (2) The improved K-means clustering algorithm and PSO algorithm are used to solve the adaptive deployment optimization problem, and the intelligent deployment and dynamic adjustment of all heterogeneous cellular network base stations and all mobile edge servers are completed to improve network service quality and computing resource utilization. The specific steps for constructing the adaptive deployment optimization problem of edge intelligent network nodes in step (1) are as follows: (1-1) Assume the system uses frequency division multiple access (FDMA) technology, macro base stations and micro base stations operate on different frequency bands, and allocate the required network resources to factory equipment through a collaborative approach. a ij =1 indicates that the i-th factory device is associated with the j-th edge intelligent network node, and vice versa, i∈[1,D]; x ik =1 indicates that the k-th subcarrier is assigned to the i-th factory device, otherwise it is not assigned, k∈[1,K]; (1-2) Construct a QoS model for a discrete manufacturing edge intelligent network system; considering the numerous obstacles in the factory, such as large metal surfaces, large equipment, and buildings, and the impact of machine operation and production activities on the transmission quality of wireless signals, the wireless channel between factory device i and edge intelligent network node j is described using the WINNER II B3 channel model; assuming the transmit power of factory device i is... The carrier frequency is f ik The distance between device i and its associated edge intelligent network node j is d. ij The path loss model is then expressed as: Where LOS and NLOS represent line-of-sight propagation and non-line-of-sight propagation, respectively, X σ Let be the loss correction factor caused by obstacles; therefore, the signal power received by edge intelligent network node j is expressed as: Affected by shadow decay, For a random variable that follows a log-normal distribution, the signal reception failure rate is defined as the received power being less than the minimum received power. The probability of that, i.e.: in, Let be the average signal power received by base station j, σ be the standard deviation of shadow fading, and Q(z) be the probability that the random variable x, which follows a standard normal distribution, is greater than z. Considering the co-channel interference from other devices using the same carrier frequency as device i within the factory, as well as the impact of mechanical vibration, electromagnetic interference, and random noise generated by equipment operation on the data transmission of device i, the uplink SINR of the i-th factory device is expressed as: in, G represents the co-channel interference experienced by device i. ijk (t): P represents the thermal noise of factory equipment. ijk (t)=A∑ l δ(t-lT) represents the noise generated by equipment operation, switching action or other periodic activities, where A is the pulse amplitude, δ(t) is the Dirac delta function, T is the pulse period, and lT is the specific time point when the l-th pulse occurs; Assume m ij b represents the data size transmitted from the i-th factory device to the j-th edge intelligent network node. ij Given the network bandwidth allocated to the i-th factory device, the sum of transmission delays for all factory devices within the coverage area of ​​the j-th edge intelligent network node is expressed as: Assuming the j-th edge intelligent network node has sufficient network capacity, the sum of the data size of all factory devices within its coverage area is expressed as: The system employs quadrature phase shift keying (QPS) modulation for signal transmission to effectively utilize limited communication bandwidth. If the m-th factory device... ij The probability of successful data transmission for each bit or data unit is the product of its success rate. Therefore, the probability that all factory equipment within the coverage area of ​​the j-th edge intelligent network node successfully transmits data is: (1-3) Constructing an adaptive deployment optimization problem for a discrete manufacturing edge intelligent network: If the key indicators of system network service quality include communication reachability, system latency, and transmission reliability, then the service quality of edge intelligent network nodes is defined as follows: Where z represents the deployment location of the edge intelligent network node, ω1,ω2,ω3∈[0,1] are the weighting coefficients of each key indicator in the system, and ω1+ω2+ω3=1, T j T is the sum of the actual data upload latency of all factory equipment within the coverage area of ​​the j-th edge intelligent network node. req The minimum latency that should be met for data transmission of factory equipment within its coverage area; R j R represents the transmission capacity that the j-th edge intelligent network node can provide. req Average network capacity required for data upload from factory equipment within the coverage area; PSR j To ensure the reliability of data transmission between factory equipment within the coverage area of ​​the j-th edge intelligent network node, PSR req The minimum reliability that must be met for successful data transmission from factory equipment within the coverage area; Considering the uneven distribution of factory equipment and communication needs, the rational and efficient deployment of edge intelligent network nodes can provide factory equipment with low latency, high reliability, and high data rate quality of service, while meeting the constraints of maximum system capacity and minimum transmission latency. Therefore, the adaptive deployment optimization problem of edge intelligent network nodes can be expressed as: C7:ω1+ω2+ω3=1 in, This is a service quality redundancy factor designed to address service fluctuations caused by increased network load. Constraint C1 states that a device will only connect to one base station at a time. Constraint C2 states that each device transmits data only on a specific sub-channel, and data transmission through multiple sub-channels is not allowed simultaneously, thus avoiding inter-channel interference and packet collisions. Constraint C3 states that the total bandwidth allocated to all devices cannot exceed the total bandwidth of the base station, and the bandwidth allocated to each device must be non-negative. Constraints C4 to C6 respectively indicate that the uplink transmission rate, latency, and reliability of each factory device i connected to the edge intelligent network node j can be basically guaranteed, where R... min T max and PSR min These represent the minimum transmission rate, maximum allowable transmission delay, and minimum reliability of factory equipment i, respectively.

2. The adaptive deployment method for edge intelligent networks for discrete manufacturing according to claim 1, characterized in that, The specific steps of step (2) are as follows: (2-1) Identify potential QoS-deficient areas By monitoring the real-time data transmission of various devices in a discrete manufacturing smart factory, and utilizing a preset QoS threshold standard R... min T max and PSR min Devices with unmet QoS requirements are detected, and then the number and spatial distribution of these devices are analyzed to quickly identify potential underserved areas. The set of devices with unmet QoS requirements is... (2-2) The improved K-means algorithm is used to perform dynamic clustering initialization on factory equipment that does not meet QoS, and the cluster center is used as the candidate deployment location of edge intelligent network nodes. Assume x j For decision variables, j indexes all possible micro base station locations, and x... j =1 indicates that a base station is deployed at location j, otherwise it is not deployed. For each factory device i∈U, a ij =1 indicates that device i, whose QoS is not satisfied, is covered by the micro base station at location j; conversely, 0 indicates that the QoS of device i is still not satisfied. If the deployment cost of each micro base station is C... j The objective function of dynamic clustering is then expressed as: For the optimization problem P2, the improved K-means algorithm is used to solve it. The number of cluster centers is dynamically adjusted in real time according to the number of devices whose QoS is not met and their spatial distribution, until an edge intelligent network node deployment scheme that can both ensure the service quality requirements of all devices in set U and maximize cost-effectiveness is found. (2-3) Use the adaptive weighted PSO algorithm to accurately adjust the location and configuration of heterogeneous cellular network base stations and MEC servers; In the PSO algorithm, each particle has a position vector X and a velocity vector V. X represents the particle's position in the solution space, and V represents the direction and velocity that determine its flight. Assuming... This represents the velocity of the i-th particle at time t. This represents the position of the i-th particle at time t. G represents the best position found in the history of the i-th particle. best This represents the best position found throughout the entire population's history. In each iteration, each particle is compared to its own historical best position. and the global best position g in the group best To update its speed and position, that is: Where r1 and r2 are random numbers, and c1 and c2 represent the acceleration weights that push the particle to its individual optimal position and the group optimal position, respectively. To maintain a balance between convergence speed and search effect, neither c1 nor c2 is 0; ω is the inertia weight coefficient at time t, which controls the continuity of particle velocity updates. A higher weight coefficient is helpful for global search, while a lower weight coefficient is helpful for local search. Fixed inertia weights cannot effectively balance exploration and utilization in complex multi-objective optimization problems, causing particles to converge to local optima too early and making it difficult to adapt to the dynamically changing factory environment. To effectively balance global and local search, the inertia weight is defined as: oh = oh max -(oh max -oh min )·(t / T) p Where, ω max and ω min represents the maximum and minimum values ​​of the inertia weight coefficient, respectively. p is a constant, and T is the total number of iterations of the algorithm. In the early stage of the search, a larger inertia weight is used to help the algorithm explore a wider solution space. In the later stage of the iteration, the inertia weight is gradually reduced, which enhances the local search ability of the particles, improves the accuracy of the optimal solution and the convergence of the algorithm.

3. An adaptive deployment system for edge intelligent networks in discrete manufacturing, operating the method described in any one of claims 1-2, characterized in that, The system comprises multiple factory devices, multiple heterogeneous cellular network base stations, multiple mobile edge computing servers, and a cloud server. The heterogeneous cellular network base stations include macro base stations and micro base stations. Micro base stations are deployed in areas where macro base station signal coverage is weak or difficult to achieve, and are connected to macro base stations via wireless fronthaul technology to enhance network signal quality and coverage. Each factory device connects to an edge intelligent network node that provides the maximum signal-to-interference-plus-noise ratio based on its actual communication environment, ensuring that the device always maintains optimal network connection quality. Each macro base station connects to a mobile edge server, which is responsible for local data processing and computing tasks, reducing data transmission latency and improving edge computing efficiency. Ultimately, the cloud server enables the coordination and optimized management of global data.

Images